Machine learning to find relevant information
Machine learning algorithms simulate how a human brain processes information, on a digital computer, allowing for artificial intelligence systems to sort through massive amounts of data and automating manual tasks. Machine learning, combined with natural language processing, can be used to rapidly process through text-based data such as scientific articles, to sort out what is relevant to the user and extract desired information.
Working with a data science non-profit volunteer organization, DataKind, and Conservation International, the SNAPP Evidence-Based Conservation Working group developed an open access machine learning application, Colandr, that aims to optimize the process of finding and extraction desired information from scientific publications in conservation research topics. You can learn more and join the community at ColandrCommunity.com.
Current projects aim to evaluate performance of this platform and provide guidance for future development. Please contact us if you are interested to help us test out the app!
Technology and finding synthesized evidence
Data visualization is increasingly widely used to allow users to dynamic explore data for multiple purposes, from learning and hypothesis exploration, to detecting patterns and validating theories (great synopsis in Harvard Business Review). With the level of complex social and ecological data captured in a typical environmental systematic map or review, delving into the different layers and iterations of the data based on how you filter the information – would be impossible in static representation (e.g. PDFs). Data visualizations allow for dynamic exploration of information – a boon to increasing accessibility and intelligibility of data from systematic evidence synthesis. In response to stakeholder need, the SNAPP Evidence-Based Conservation group developed and maintain the Evidence for Nature and People Data Portal to visualize data from systematic maps and reviews on linked social and ecological outcomes.
Knowledge uptake and learning outcomes from data portals and knowledge management platforms: There has been a recent boom of data platforms that intend to improve access and openness of data sets important for understanding ecosystems and human communities. However, there is little understood about how the design and communication of platforms can help or hinder learning and use of knowledge provided on these sites. We are currently investigating how we can formulate a theory-based evaluation approach to understanding learning and uptake.